
GEOPHYSICAL RESEARCH LETTERS, VOL. 33, L01301, doi:10.1029/2005GL024047, 2006 Plume-lithosphere interaction beneath a fast moving plate Catherine Thoraval,1 Andre´a Tommasi,1 and Marie-Pierre Doin2 Received 11 July 2005; revised 12 October 2005; accepted 25 October 2005; published 5 January 2006. [1] Two-dimensional numerical simulations of mantle the plume [Von Herzen et al., 1989]. A small increase of convection with temperature and pressure dependent surface heat flow (5–10 m WmÀ2) is nevertheless observed viscosity are used to study plume–lithosphere interaction below the Cabo Verde, Bermuda, and La Reunion hotspot beneath a fast moving plate. Plumes behavior and, hence, tracks, suggesting some lithospheric thinning downstream of their erosional and melting potential, depend on the the actual hotspot location [Bonneville et al., 1997]. Rayleigh number and plume buoyancy flux. Analysis of [4] 2D and 3D numerical models performed under the balance between large-scale and plume-induced flow steady-state or quasi steady-state conditions show that and of the ability of the plume to melt allows to put bounds thermo-mechanical erosion of the lithosphere above a on the upper mantle viscosity (1020,1021 Pas), on the mantle plume is a slow process and hence turn out plumes diameter (<200 km) and temperature anomaly significant reheating of the lithosphere beneath a moving (200–400°C). Within this range of parameters, strongly plate [Monnereau et al., 1993; Davies, 1994; Ribe and time-dependent small-scale instabilities form in the plume- Christensen, 1994, 1999; Sleep, 1994]. In contrast, recent lithosphere boundary layer. They lead to thermal rejuvenation 3D numerical models suggest that the strongly time of the lithosphere downstream from the plume. The 1200°C dependent small-scale convective instabilities in the low- isotherm is raised by up to 30km, but the 800°C isotherm is viscosity layer formed by the spreading of hot plume hardly moved, leading to a steep transient geotherm at the material at the base of the lithosphere may be an effective base of the plate. Citation: Thoraval, C., A. Tommasi, and M.-P. mechanism to erode the base of the lithosphere, even for a Doin (2006), Plume-lithosphere interaction beneath a fast moving fast-moving plate like the Pacific [Moore et al., 1998]. plate, Geophys. Res. Lett., 33, L01301, doi:10.1029/ However, viscosity in these models shows a much lower 2005GL024047. temperature-dependence than the one expected in upper mantle based on laboratory deformation experiments [Karato and Wu, 1993], leading to underestimated litho- 1. Introduction sphere viscosities. [2] Mantle plumes are assumed to be thermal and/or [5] In this paper, 2D Cartesian numerical convection chemical instabilities ascending through the convecting models are used to further study the plume-lithosphere mantle. The impact of a plume head beneath the lithosphere interaction beneath a fast-moving plate. Convective desta- is generally thought to result in intense melting of plume bilisation of the lithosphere by the plume and, hence, the material, producing large igneous provinces. In a later stage, erosion of the lithosphere, are described as a function of the decompression melting in the plume tail results in hotspot plume buoyancy flux and of the global Rayleigh number. volcanic chains as the lithosphere moves above it. Despite The ability of modelled plumes to produce partial melting is their numerous signatures at the surface, the processes used to discriminate Earth-like plumes. taking place at the plume-lithosphere boundary layer are still poorly understood. A major open question regards the ability of a plume to significantly erode the lithosphere. 2. Model 3 [ ] Observations of the lithosphere thickness and erosion [6] Small-scale convection is a highly time-dependent atop mantle plumes are highly controversial. Seismic feature, associated with strong viscosity gradients, which studies beneath Hawaii lead to opposite conclusions. description requires time-costly models with very fine grids. Surface-wave dispersion as well as sP converted waves Thus to investigate a large range of Rayleigh numbers and data require no significant thinning, but very low velocities plume buoyancy fluxes, we performed 25 numerical experi- in the asthenosphere [Bock, 1991; Woods et al., 1991; ments using a 2D convection code [Christensen, 1983, Woods and Okal, 1996; Priestley and Tilmann, 1999]. On 1984]. the other hand, recent S-wave receiver function data image [7] The modeled domain has open bottom and down- a gradual thinning of the lithosphere from the present stream side boundaries. A velocity field is imposed on top to hotspot location, beneath Hawaii, towards Kauai, where mimic plate motion (the ridge is located at the upper-left the lithosphere is reduced by half [Li et al., 2004]. Heat corner). Constant temperature is imposed on top and bottom flow data also lead to contradictory conclusions. Comparison sides. Internal heating is not taken into account. The of on-swell and off-swell data for the Hawaii hotspot points to numerical grid consists of rectangular elements of variable no significant thermal rejuvenation of the lithosphere above size. The grid is refined horizontally where the plume is to be introduced and vertically at the top of the box to avoid 1Laboratoire de Tectonophysique, Universite´ Montpellier 2, Montpel- numerical artifacts and to achieve a better description of the lier, France. plume-lithosphere interaction. The calculations are performed 2Laboratoire de Ge´ologie, Ecole Normale Supe´rieure, Paris, France. in a non-dimensional space. The re-dimensionalisation Copyright 2006 by the American Geophysical Union. scheme is based on the box thickness and on the temper- 0094-8276/06/2005GL024047$05.00 ature drop from the bottom to the top of the box (fixed to L01301 1of4 L01301 THORAVAL ET AL.: PLUME-LITHOSPHERE INTERACTION L01301 hotspots in the South Pacific Superswell might come from the transition zone, in relation with the large region of slow seismic velocities extending throughout the lower mantle [Davaille, 1999]. [11] Each plume is characterized by its maximum tem- perature anomaly DT and by its diameter Ø, defined by the position at which the temperature anomaly is reduced by a factor e. Diameters range from 35 km to 200 km. Recent magnetotelluric data require that the radius of the Hawaii plume is less than 100 km [Constable and Heinson, 2004]. A maximum plume radius of 70 km is also deduced from the sharpness of the bend in the Hawaiian-Emperor hotspot track [Duncan and Richards, 1991]. In contrast, recent seismic tomography images of mantle plumes show con- duits with radius between 100–400 km extending through- out the mantle [Montelli et al., 2004]. Temperature anomalies vary between 100°C and 400°C. This tempera- ture range is consistent with values derived from the Figure 1. Regime diagram describing the plume behavior variation in the thickness of the mantle transition zone as a function of the Rayleigh number and of the 2D plume below the Society or Galapagos mantle hotspots [Niu et buoyancy flux. Gray symbols mark plumes that produce al., 2002; Hooft et al., 2003]. anhydrous melting. Labels identify the models presented in [12] Melting has been simply addressed by delimiting the Figures 2 and 3. regions where the solidus temperature is exceeded. We used the melting curves given by Katz et al. [2003] for anhydrous 2 melting (Tsolidus,°C = 1085.7 + 132.9 PGpa À 5.1 PGpa). H = 700 km and DTB = 1350°C respectively). The aspect ratio of the box is 1/10. The age of the lithosphere at the 3. Results upper right corner is ca. 95 m.y. for a plate velocity [13] The behavior of plumes depends both on Ra and on of 13.5 cm/y. The minimum grid step for most models the plumeZ buoyancy flux, which in 2D may be estimated as is <7 km, but resolution tests have also been performed with a minimum grid step of 4 km. B = arDTvz dx, where W is the bottom of the model [8] The viscosity is Newtonian and depends exponentially W on temperature and pressure according to the following (Figure 1). At low Ra, small plumes are entrained by the EaþVaP large-scale flow induced by plate motion and cannot reach equation n = n eðÞRT . The activation energy Ea and the 0 the base of the lithosphere. When B is increased, plume activation volume Va have been set to 250 kJ/mol and material can ascend, but the conduit is deflected at depth by 2.25 cm3/mol, n is a reference viscosity and R is the gas 0 the large-scale flow and the plume reaches the basis of constant. Rayleigh number (Ra) varies from 2.105 to 1.107, the lithosphere downstream from its original position corresponding to viscosities at the bottom of the box (nB) argDT H 3 (Figure 2a). At higher Ra, plumes are little perturbed by of 4.1019 and 2.1021 Pa s according to Ra = B where the large-scale flow and ascend easily throughout the upper knB mantle (Figure 2b). When the plume reaches the base of the a is the thermal expansivity, r the density and k the thermal lithosphere, most of the hot plume material flows down- diffusivity respectively set to 3.10À5 KÀ1, 3300 kg/m3 and stream, entrained by the plate motion, and a plume- 1.10À6 m2/s. The specific heat is set to 1250 J/kg K. [9] Initial conditions are obtained by running the model without plume until equilibrium is reached. The plate motion is responsible for a large-scale flow. The lithosphere first cools following the half-space model. After an onset time, function of Ra, small-scale convection develops, leading to a constant lithosphere thickness at old ages [Davaille and Jaupart, 1994; Dumoulin et al., 2001].
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages4 Page
-
File Size-